Process and device for bite-rate reduction for the recording of images on a video recorder

Abstract

A method comprising breaking an image down into sub-bands, performing a pre-analysis of the image in image blocks, recursively and adaptively assigning the available bit-rate to each pixel in the sub-bands, selecting a quantizer determined by the source probability density and a calculation of the minimum quantizing error variance, and coding each pixel of each signal component in every sub-band according to the selected quantizer. The method is applicable to HDTV digital VCRs.

We claim: 
     
       1. Method of bit-rate reduction for recording images on a digital video recorder, the method comprising the steps of: splitting a source image into plural sub-bands,   pre-analyzing the source image by image blocks,   allocating an available bit rate to each pixel of the plural sub-bands recursively and adaptively, as a function of a spatial parameter and a frequency parameter, the spatial parameter calculated during the pre-analysis step,   selecting a quantizer determined by a probability density of the source image and a minimum variance calculation for a quantization error, and   coding, as a function of the selected quantizer, each pixel of each component of the source image in each of the plural sub-bands, wherein the step of pre-analyzing the source image by image blocks comprises the sub-step of     calculating an adaptive weighting coefficient, cod(k,n), based on a difference in each of the plural sub-bands between values, F p  (k,i,n), of each pixel of a block and a first variance, Var(i,n), of the each of the plural sub-bands in accordance with the relation ##EQU15## where i is a sub-band number, n represents a luminance component or a chrominance component, k denotes a block number, N is a number of sub-bands used, F p  (k,i,n) represents a value of the pixel belonging to block k of sub-band i for component n, Var(i,n) is the first variance of sub-band i for component n, cod(k,n) is an adaptive weighting coefficient for block k of component n, and α is a coefficient dependent on the luminance and chrominance components.   
     
     
       2. Method according to claim 1, wherein the step of allocating the available bit rate to each pixel of each sub-band comprises the sub-steps of: allocating bits to lower frequency bands to a greater extent than higher frequency bands, and   weighting each image block by the weighting coefficient cod(k,n); and   wherein the step of coding comprises coding each pixel using b(k,i,n) bits.   
     
     
       3. Method according to claim 2, wherein the step of allocating the available bit rate b(k,i,n) for each pixel of each sub-band comprises the steps of: determining a coefficient λ determined by using Lagrange multipliers in order to minimize the variance of the quantization error in each sub-band,   allocating the available bit rate based on a variance of each component n of a band i weighted by a square of the adaptive weighting coefficient cod(k,n) and divided by a number of pixels, N k , per band, and   weighting the variance by the coefficient λ determined using the Lagrange multipliers.   
     
     
       4. Method according to any one of claims 1, 2 and 3, further comprising the step of recording an image sub-sampled from the source image compatibly with existing digital video recorders. 
     
     
       5. System for reducing a bit-rate for recording images on a digital video recorder, the system comprising: a sub-band splitter for splitting an image into plural sub-bands,   a linear low frequency quantizer coupled to the sub-band splitter,   a non-linear high frequency quantizer,   a multiplexer;   a sub-band variance calculating device;   a device for measuring energy per block;   a calculating device for calculating a number of output levels of the linear low frequency and non-linear high frequency quantizers, the calculating device being controlled by the sub-band variance calculating device and the device for measuring energy per block; and   a pre-analyzer for calculating an adaptive weighting coefficient, cod(k,n), based on a difference in each of the plural sub-bands between values, F p  (k,i,n), of each pixel of a block and a first variance, Var(i,n), of the each of the plural sub-bands in accordance with the relation ##EQU16## where i is a sub-band number, n represents a luminance component or a chrominance component, k denotes a block number, N is a number of sub-bands used, F p  (k,i,n) represents a value of the pixel belonging to block k of sub-band i for component n, Var(i,n) is the first variance of sub-band i for component n, cod(k,n) is an adaptive weighting coefficient for block k of component n, and α is a coefficient dependent on the luminance and chrominance components.   
     
     
       6. Method according to any one of claims 1, 2 and 3, further comprising the step of recording the plural sub-bands compatibly with existing digital video recorders.